The frontier between testing and targeting is not always clearly defined, and the decision to choose one or the other can get confusing. To make matters more complicated, there are several targeting approaches to choose from, including behavioral targeting and personalization. As a digital marketer, should you care about the differences between each technique? Are there specific cases where one technique is more efficient than the other?
Let’s take a closer look at the actual differences and identify some opportunities that lie within each approach.
What Is Behavioral Targeting?
In its simplest form, targeting can be thought of as a subset of testing, where an experiment is conducted for a specific audience of your site. Examples include testing an internal search results page for paid search traffic or testing a product page for your email marketing campaigns.
Behavioral targeting leverages data that you own about your users, which can be used as predictors for future behavior. Personalizing the online experience for a subset of your users based on previous shopping behavior is a typical example of behavioral targeting.
Where Should You Start?
The answer to this question depends on a number of factors, including your own testing maturity. Let’s look at a few considerations that should help answer the question.
Sample Size Implications
If you find it challenging to get a large enough sample size to reach statistical significance and stabilization, keep in mind that targeting focuses on subsets of your user base, which will in turn translate into smaller sample sizes. Furthermore, depending on your traffic assignment strategy, running targeted tests might lower traffic to other experiments.
That being said, because targeted tests cater to groups of users who generally share similar behavior, it is not uncommon for this type of testing to reach statistical significance faster than non-targeted tests.
Critical Segments Behavior
Existing test data is a great source for identifying targeting opportunities. Try to focus on your key segments and apply them to existing test data. Differences in behavior might not always show up for your most critical metrics, but can appear for different areas of your website. Critical segments of traffic that behave similarly in response to specific tests are great targeting prospects.
Opportunities for targeting don’t always surface from deep-dive analysis of your web analytics and test data. It might very well be that you have content that caters to specific audiences. Do you have tailored offers for particular geographical regions? Does your website offer different types of membership or you do you have premium and standard customers? Similar scenarios are natural candidates for targeted testing.
The Big Picture
By narrowing your focus on specific groups of users, you can miss effects that can impact the larger population. Remember that you can always segment results of an experiment and analyze the behavior of different segments independently. On the other hand, once an experiment is targeted to a specific group of users, you cannot expand the results to your entire population.
At the end of the day, there is no right or wrong answer. If you are just getting started with A/B or multivariate testing, targeting might not be the right place to start. As you start learning more about your users through an iterative testing approach, you will find that segments of users naturally emerge from your test data and can easily translate into successful targeting test scenarios.